How to Make an Artificial Intelligence Algorithm "Ecological"? Insights from a holistic perspective.

Nowadays, Artificial Intelligence is growing in many daily activities. On the one hand, it has many positive effects and produces social benefits. On the other hand, its development and deployment raise issues related to biases, such as gender, disability, and culture. Moreover, Artificial Intelligence’s growing autonomy in decision-making could lead to decisions that conflict with human values or harm individuals and society. These issues stem from biased or incomplete datasets and a lack of transparency and accountability in the algorithms. Consequently, paying increasing attention to the ongoing discourse on Artificial Intelligence ethics: its autonomy in decision-making, and biases is necessary. A human-centric approach is a minimum requirement for designing algorithms since this approach is aligned with human values, dignity, and goals. Notwithstanding, its application does not guarantee a deep understanding of the context of use. According to recent theoretical perspectives, a deep interpretation of the context of use (i.e., a holistic perspective) could better regulate ethical aspects. This paper goes in this direction, presenting a human-centric and ecological approach as a design methodology. It has been experienced within Use Case 6 of the European FRACTAL project, which aims to develop intelligent totems for advertising and customer assistance in sentient shopping malls. The intelligence is realized by several artificial intelligent algorithms (e.g., gender recognition algorithms). By adopting Bronfenbrenner’s ecological approach, algorithms were made free from gender bias, mirroring the context of men’s and women’s use at shopping malls as it is currently, i.e., characterized by gender balance. This proposal contributes to the ongoing discourse on Artificial Intelligence ethics and the development of its ethical algorithms.